From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ben Greear Subject: Re: [PATCH] veth: remove hardware checksum feature Date: Thu, 08 Aug 2013 12:43:34 -0700 Message-ID: <5203F4E6.1010701@candelatech.com> References: <51F15E50.8080208@guap.ru> <5202E153.4060202@candelatech.com> <1375920752.4004.71.camel@edumazet-glaptop> <5202E510.9060309@candelatech.com> <1375924125.4004.83.camel@edumazet-glaptop> <5202FA51.4040906@candelatech.com> <1375930325.4004.88.camel@edumazet-glaptop> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org To: Eric Dumazet Return-path: Received: from mail.candelatech.com ([208.74.158.172]:39648 "EHLO ns3.lanforge.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S965983Ab3HHTnf (ORCPT ); Thu, 8 Aug 2013 15:43:35 -0400 In-Reply-To: <1375930325.4004.88.camel@edumazet-glaptop> Sender: netdev-owner@vger.kernel.org List-ID: On 08/07/2013 07:52 PM, Eric Dumazet wrote: > On Wed, 2013-08-07 at 18:54 -0700, Ben Greear wrote: > >> Maybe we should just do the csum calc in the kernel if packet is >> about to be sent up to user-space via af_packet? I think that >> would keep the expected behaviour, and hopefully not loose any of the performance >> benefits for cases where the packet never leaves the kernel? > > I think you are rephrasing what I suggested ;) > > I'll send a patch asap, unless someone beats me. I'll be happy to test. And a slightly related question if you have the time: My kernel-mode bridge was breaking as well, but this appears to be a bug in my code. I had this code (since 2009, at least): // Evidently this fixes issues with sending between NICs // that support and do not support hw-csum. skb->ip_summed = CHECKSUM_NONE; It seems that at some point, this fixed some problem I saw, but it was long ago. If I remove this, then the 3.9 kernel bridges just fine between the VETH and a physical interface. I do not see any modification of the ip_summed in the bridge code. So, the question is: Is there any time that I *should* be mucking with skb->ip_summed when bridging pkts from one device to another on modern kernels? I'm using the ptype_all hook to grab packets, and ndo_start_xmit to send them, in case that matters. Thanks, Ben -- Ben Greear Candela Technologies Inc http://www.candelatech.com